"The Promise of Watermarking AI Content"

A group of major technology companies recently made an announcement about watermarking Artificial Intelligence (AI)-generated content. However, cybersecurity researchers already suggest this new approach has flaws. Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI met with the White House to discuss how they can mitigate the risks associated with the AI they create. They promised to invest in cybersecurity and AI-generated content watermarking. Dr. Florian Kerschbaum, a professor of computer science and member of the Cybersecurity and Privacy Institute at the University of Waterloo, explained that the watermarking technology pitched by the companies embeds a secret message within the content's code. The idea is that the message cannot be removed without also removing the content. Kerschbaum notes, however, that there are still some uncertainties in the scientific foundations of watermarking. Malicious actors may be able to remove a watermark, and the issue of digital watermarks has fascinated scientists for decades. This article continues to discuss concerns and questions regarding watermarking AI-generated content.

The University of Waterloo reports "The Promise of Watermarking AI Content"